Reading the Mix in Practice
No large operator runs a single mode. The right question is not build versus lease in the abstract but what blend a given company runs, in which markets, and how that blend is changing. The mix is the signal.
Start with the observable spread. Among the largest US operators, the pattern over the past several years has been roughly this: the biggest self-builders, Google and Meta, own a large majority of their footprint and lease at the margin to enter new regions fast or to bridge a construction gap. Microsoft and Amazon have historically run more balanced books, owning core campuses while leasing heavily from wholesale developers to keep pace with demand they could not pour concrete fast enough to serve. Newer entrants and enterprises with smaller footprints lean on colocation and wholesale leasing because they lack the scale to justify a self-build program.
Those blends are not fixed. Watch how they move. When a historically build-heavy operator suddenly signs a wave of wholesale leases, one of two things is usually happening: demand has outrun its ability to build, so it is renting capacity to avoid losing workloads to a competitor, or it is testing a new market before committing owned capital. Both are bullish for near-term demand. Conversely, when a lease-heavy operator begins buying land and announcing self-builds in a region, it is signaling that it now views that region's demand as durable enough to own, a stronger and longer-dated signal.
The mix also encodes a speed-versus-margin trade. Leasing is faster: a developer with land already entitled and power already contracted can deliver a shell in half the time an operator starting from raw land would take. But leasing carries the developer's return on top of construction cost, so over a fifteen-year horizon it is more expensive per megawatt than owning. An operator that tilts toward leasing is, in effect, buying time and paying for it. When you see that tilt, ask what the operator is racing against. Usually the answer is a demand surge it did not fully anticipate, which is precisely the kind of signal that matters.
Geography sharpens the read. A company might self-build in its established heartlands, central Ohio, northern Virginia, the Nordics, where it has power relationships and construction crews in place, while leasing in newer or harder markets where entitlement and power are slow. So the same operator can be build-heavy in one metro and lease-heavy in another, and the difference maps to where it has infrastructure muscle versus where it is still buying its way in. A shift from leasing to building in a specific metro is a metro-level conviction signal worth more than any national average.
There is a financing dimension too. The past two years have seen a surge in off-balance-sheet structures, joint ventures and build-to-suit vehicles that let an operator control a facility economically while keeping the construction debt off its own books. When you see a hyperscaler forming a multibillion-dollar JV with an infrastructure fund, read it as a build in economic substance even though it appears in the lease or partnership column. The accounting form understates the conviction.
Practically, to read a company's mix, triangulate three sources: capital-expenditure disclosures and the split between purchases of property versus finance-lease additions in the cash-flow statement; announced land acquisitions and construction permits, which precede owned builds by quarters; and wholesale developers' leasing announcements and backlog, which name their anchor tenants. Put those together and the blend, and its direction, comes into focus. The direction of the blend is usually a better forward indicator than the level, because it captures the operator responding in real time to demand it is seeing before the rest of the market does.